MENTAL MODEL · M3716

GANs

DecideHigh supportGame Theory

Included

Version 1.0.0 · Updated 2026-07-30

CORE DEFINITION

Composed of two networks: the generator (counterfeiter) tries to produce realistic data to fool the discriminator; the discriminator (bill validator) tries to distinguish real from fake. They evolve through competition, and eventually the generator can produce data that is indistinguishable from real data.

SCAFFOLDING EFFECT

psychology

Reduce cognitive load

Left-right hand combat. The best way to improve yourself is to split into two selves: one responsible for doing (generator) and one for criticizing (discriminator). In a team, set up a dedicated 'blue team' or 'red team' to attack your own proposals.

anchor

Anchor fast decisions

A GAN consists of two neural networks trained adversarially: the generator creates fake samples, the discriminator distinguishes real from fake, and their competition continuously improves the quality of generation. Proposed by Goodfellow in 2014.

MINIMUM ACTION

In progress 0/1

Practice this model in one real situation:

Build a generator and a discriminator, and define an adversarial loss. Train both alternately until near Nash equilibrium. Evaluate using the discriminator's convergence.

Check to track your progress (stored locally)

Learning progress0%

account_treeGenealogyexpand_more

menu_bookReferencesexpand_more

Source support: Explicit

link
zh.wikipedia.orghttps://zh.wikipedia.org/wiki/%E7%94%9F%E6%88%90%E5%AF%B9%E6%8A%97%E7%BD%91%E7%BB%9CZH · Explicit
verified

RELATED MODELS